Smart Detection of Elephant Intrusions Using LoRaWAN Enabled IoT Sensors

Abstract Despite recent progress in mitigating human elephant conflict across Sri Lanka, it continues to pose severe threats to human lives, security and elephant conservation, not only in Sri Lanka but also in other parts of Asia and Africa. Limited coverage, unreliable sensors, on-site maintenance, and lack of a communications infrastructure in remote areas are all limitations of current available mitigation technology (e.g., fences, manual guarding, single sensor alarms). In this paper, a smart intrusion detection system is designed for elephants which integrates both seismic and passive infrared sensors, locally processes the data, uses a machine learning algorithm to classify the event, communicates via LoRaWAN, sends the alerts to the cloud and includes a non-lethal deterrent counter measure. The proposed system has four coordinated layers: multi sensor field nodes, a hybrid LoRaWAN communication backbone to connect to the WiFi and LTE networks as a failover option, a machine learning engine to discriminate between elephants and non-elephants, and last but not least, a dashboard and notification layer for farmers and wildlife officers. The architecture has been developed with these features – low power, self-healing communication, offline buffering, delivery of warning in real time. The paper also implies the design on the backdrop of the latest literature from the open sources for human elephant conflict, seismic elephant detection and LoRaWAN for wildlife monitoring. In a recent geophone interface study done in Sri Lanka, up to 155.6 m distance was achieved for seismic detection and an accuracy of 99.5% was achieved which provides strong support to the sensing basis of this project. The system being proposed further builds on that as a full farm platform for deployment in the rural environment.